Non-Coding Cis-Element of Period2 Is Essential for Maintaining Organismal Circadian Behaviour and Body Temperature Rhythmicity

ARTICLE

https://doi.org/10.1038/s41467-019-10532-2 OPEN Non-coding cis-element of Period2 is essential for maintaining organismal circadian behaviour and body temperature rhythmicity

Masao Doi 1,7,8, Hiroyuki Shimatani1,7, Yuta Atobe1,7, Iori Murai1,2, Hida Hayashi1, Yukari Takahashi1, Jean-Michel Fustin 1, Yoshiaki Yamaguchi1, Hiroshi Kiyonari3, Nobuya Koike4, Kazuhiro Yagita4, Choogon Lee5, Manabu Abe6, Kenji Sakimura6 & Hitoshi Okamura1,2,8 1234567890():,; Non-coding cis-regulatory elements are essential determinants of development, but their exact impacts on behavior and physiology in adults remain elusive. Cis-element-based transcriptional regulation is believed to be crucial for generating circadian rhythms in behavior and physiology. However, genetic evidence supporting this model is based on mutations in the protein-coding sequences of clock genes. Here, we report generation of mutant mice carrying a mutation only at the E′-box cis-element in the promoter region of the core clock gene Per2. The Per2 E′-box mutation abolishes sustainable molecular clock oscillations and renders circadian locomotor activity and body temperature rhythms unstable. Without the E′-box, Per2 messenger RNA and protein expression remain at mid-to-high levels. Our work delineates the Per2 E′-box as a critical nodal element for keeping sustainable cell-autonomous circadian oscillation and reveals the extent of the impact of the non-coding cis-element in daily maintenance of animal locomotor activity and body temperature rhythmicity.

1 Department of Systems Biology, Graduate School of Pharmaceutical Sciences, Kyoto University, Sakyō-ku, Kyoto 606-8501, Japan. 2 Laboratory of Molecular Brain Science, Graduate School of Pharmaceutical Sciences, Kyoto University, Sakyō-ku, Kyoto 606-8501, Japan. 3 Laboratories for Animal Resource Development and Genetic Engineering, RIKEN Center for Biosystems Dynamics Research, Kobe 650-0047, Japan. 4 Department of Physiology and Systems Bioscience, Kyoto Prefectural University of Medicine, Kyoto 602-8566, Japan. 5 Department of Biomedical Sciences, College of Medicine, Florida State University, Tallahassee FL 32306, USA. 6 Department of Cellular Neurobiology, Brain Research Institute, Niigata University, Niigata 951-8585, Japan. 7These authors contributed equally: Masao Doi, Hiroyuki Shimatani, Yuta Atobe. 8These authors jointly supervised this work: Masao Doi, Hitoshi Okamura. Correspondence and requests for materials should be addressed to M.D. (email: [email protected]) or to H.O. (email: [email protected])

NATURE COMMUNICATIONS | (2019) 10:2563 | https://doi.org/10.1038/s41467-019-10532-2 | www.nature.com/naturecommunications 1 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-10532-2

vidence shows that non-coding cis-regulatory elements are degree of the impact of the deletion of the Per2 E′-box on the as important as protein-coding sequences for determining organismal clock: The Per2 E′-box is essential for the period E 1–3 cell identity and morphological development . The degree determination of behavioral rhythms in DD conditions and for of the effects of non-coding cis-regulatory elements on animal sustaining stable rhythms under LL and jet-lag conditions. These behavior and physiology after development, however, remains data underscore the roles of the non-coding cis-element in the elusive. Circadian clocks generate ~24 h rhythms in behavior and regulation of daily behavior and physiology in adulthood. physiology, which allow organisms to anticipate and adjust to daily environmental changes. The rhythm generating mechanism Results of the circadian clock involves clock genes, which regulate their Generation of Per2 E′-box mutant mice. We developed mutant own transcription in a negative transcription-translation feedback – mice harboring a targeted mutation in only the Per2 E′-box. To loop4 6. In this canonical feedback model, non-coding cis-ele- – do this, we used the piggyBac (PB) transposase tool for genome ments are fundamental to both initiating and closing the loop7 9. engineering because it allows for seamless removal of the PB- Genetic evidence supporting this model, however, is still exclu- flanked marker cassette (PB-Neo) from the host genome after the sively based on the effects of mutations in the protein-coding mutation is introduced (Fig. 1a). Conventional methods that rely sequences of the clock genes. Therefore, while it is tempting to on the Cre/loxP or Flp/FRT system leave behind a single loxP- or speculate that the circadian clock-related non-coding cis-elements FRT-derived ectopic sequence after marker excision. Such a are operational for daily dynamic regulation of behavior and footprint sequence could affect promoter architecture and/or physiology, there is currently no direct evidence to corroborate enhancer communication of the target gene20. The PB transpo- this notion. sase (PBase)/PB system circumvents this problem21 (see Notwithstanding the recent intriguing discoveries of circadian Fig. 1b–d). Southern blot analysis confirmed that the PB-neo- oscillations in peroxiredoxin superoxidation in transcriptionally mycin cassette was deleted without re-integration into the host incompetent anucleate erythrocytes10 and the expression of such genome (Fig. 1c). The footprint-free disappearance of the marker cycles in the neurons in the suprachiasmatic nucleus (SCN)11, the cassette was further confirmed by DNA sequencing (Fig. 1d). To consensus conjecture regarding the mammalian clockwork sup- reduce the confounding effects of a mixed genetic background, ports the transcriptional feedback model. However, the literature the mutant mice were backcrossed with C57BL/6J mice until addressing mutations or deletions in the protein-coding sequen- microsatellite markers covering all individual chromosomes were ces of the clock genes cannot negate the possible contribution of congenic to the C57BL/6J strain (see Supplementary Fig. 2). The post-translational (or non-genomic) clock mechanisms. More- targeted mutation of the E′-box sequence was finally verified by over, the extent of the impact of the cis-regulated transcriptional DNA sequencing using homozygous mutant mice (Fig. 1d). feedback cycle on establishing circadian rhythmicity in vivo Chromatin immunoprecipitation (ChIP) assays using homo- remains the subject of considerable debate4,12, due in large part to zygous mutant mouse liver, sampled at 4 h intervals for 24 h the lack of genetic research on non-coding cis-element mutations. under DD conditions, revealed that the clock proteins PER1, In the current study, we develop mutant mice carrying a CRY1, CRY2, and CLOCK had consistently low binding levels to mutation in a circadian cis-acting element of the core clock gene the mutated Per2 promoter (Fig. 1e). In a control experiment with Per2. Although Per2 was cloned as a secondary period gene in WT (+/+) samples, the peak binding levels of PER1, CRY1, mammals, gene knockout studies revealed that Per2 mutant mice CRY2, and CLOCK to the Per2 promoter were observed at CT12, displayed a loss of circadian rhythmicity, revealing its prominent – CT4, CT20, and CT8, respectively, consistent with previous role in the mammalian molecular clockwork13 15. Moreover, reports19,22. These results confirmed that Per2 E′-box function familial advanced sleep phase syndrome in humans is attributed was abolished in the mutant mice. to a missense mutation in the Per2 gene16. The Per2 E′-box sequence (5′–CACGTT–3′) located near the putative transcription initiation site17 (–20 to –15) has been Per2 E′-box is essential to maintain cellular circadian oscilla- demonstrated to be the principal circadian cis-element that is tions. To investigate the role of the Per2 E′-box in maintaining sufficient to induce oscillating levels of reporter transcription via cellular oscillation, we generated primary fibroblast cultures from the mouse Per2 minimal promoter17,18. The importance of this WT (Per2E′+/+) and mutant (Per2E′m/m) mouse lung tissues and particular cis-element is further implied by the high degree of monitored circadian fluctuations in the PER2 protein levels over 80 DNA sequence conservation in its flanking region between h(Fig.2a). As expected, the endogenous PER2 protein in syn- humans and mice17 and by extremely enriched clock protein chronized WT cells displayed characteristic circadian oscillations in binding to this element19. Publicly available genome-wide ChIP- both abundance and electrophoretic mobility, which continued over seq data highlight the predominant peak of clock protein binding multiple cycles under constant culture conditions23 (Fig. 2a). In activity around this E′-box (see Supplementary Fig. 1). contrast, Per2E′m/m cells failed to maintain normal PER2 protein The present study was designed to investigate the role of oscillations (Fig. 2a). It is important to note that the mutant cells this unique E′-box sequence of the Per2 promoter (hereafter, Per2 could competently form the first surge of PER2 expression after E′-box) as a potential nodal cis-element in the mammalian clock- synchronization (4–12 h). However, following a subsequent increase work. By introducing a site-specificmutationatthiselement (28 h), PER2 expression in the mutant cells remained at mid-to- in vivo, we provide empirical evidence to show that the Per2 E′-box high levels and eventually lost apparent circadian variation after 60 is essential for maintaining cell-autonomous circadian oscilla- h (see also Supplementary Fig. 3). Electrophoretic migration tions. The cells without the Per2 E′-box cannot maintain circa- rhythms also disappeared from the mutant cells (Fig. 2a). The dian molecular oscillations in culture conditions. At the relative intensities of the three major bands of the PER2 protein organismal level, mice lacking the Per2 promoter E′-box show remained unchanged in the Per2E′m/m cells. The post-translational destabilized locomotor activity and body temperature rhythms rhythms were thus affected either directly or indirectly by the under altered light conditions, including constant light (LL) and mutation of the E′-box. We experimentally confirmed that the experimental jet-lag conditions. Due to compensatory mechan- protein-coding sequences of PER2 and known clock protein kinases isms in vivo, the mutant mice kept under constant dark (DD) and phosphatases were unaffected in our mutant mice. These conditions remain rhythmic but exhibit considerably shorter results indicate that the E′-box is indispensable for maintaining circadian periods than WT mice. Our data therefore define the normal PER2 protein oscillations.

2 NATURE COMMUNICATIONS | (2019) 10:2563 | https://doi.org/10.1038/s41467-019-10532-2 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-10532-2 ARTICLE

a b c Post- e Per2 crossing PER1-ChIP 22.0 kb Rosa 5 −20 −15 WT allele +/PB-Neo PBase +/+ PBase 4 m/m CACGTT t +600 +603 Targeted allele u 3 ATG p n I WT (+) TTAA PB-Neo 2 F1 +/+ +/PB-Neo+/m +/m* % allele E1 E2 kb 1 22.0 AseI AseI 0 3′probe 14.6 0 4 8 12 16 20 Per2 probe +/m ′ CT

0.6 kb 3

20 CRY1-ChIP 14.6 kb 14.6 18 +/+ Targeted GCTAGT AseI ATG 16 m/m t 14 u 12 PB allele TTAA TTAA Backcrossing p n 10 I (C57BL6) 8 E1 PB-Neo cassette E2 % 6 +/m 4 AseI 3′probe AseI 2 5′TRNeo 3′TR 0 0 4 8 12 16 20 probe Neo probe CT GCTAGT ATG Per2 TTAA 3.5 CRY2-ChIP Mut (m) m/m +/+ 3.5 E2 PBase 3.0 +/+

allele E1 t 2.5

u m/m

Neo p 2.0 n

AseI ′ AseI I 3 probe 1.5

% 1.0 0.5 Transcription start site Transposon-excised site 0 −20 −15 0 4 8 12 16 20 d ′ −1 +600 +603 −20 −15 E -box TTAA m/m CT G C G C G C G G TTC A CCCCG T T T AAT TTGGA C A G C GGA GGGC .... GGGAAA G G T T A A AC AG A G AA CLOCK-ChIP G C GGG T C T A G T T T CCA +/+ 6 +/+ 5 m/m 4 3 G C G C G C G G TTC A CCCCG T T T AAT TTGGA C A G C GGA GGGC .... GGGAAA G G T T A A AC AG A G AA

% Input 2 G C T A G +/m Homozygous mutation 1 0 CACGTT GCTAGT 0 4 8 12 16 20 CT

Fig. 1 Generation of mice carrying a targeted mutation at Per2 E′-box. a Genome modification strategy using the piggyBac transposon. The CACGTT Per2 E′-box was mutated to GCTAGT. Top line, genome architecture of the mouse Per2 gene; E1, exon 1; E2, exon 2; Blue, wildtype E′-box; Red, mutated E ′-box; Green, piggyBac transposon carrying a neomycin-resistant gene (PB-Neo cassette) inserted at a genomic TTAA site (+600 to +603); Gray bars, Southern blot probes. Numbering shows the position relative to the putative transcription start site (+1). b Interbreeding scheme. Mice heterozygous for the targeted allele (+/PB-Neo) were intercrossed with ROSA26-PBase mice. The resultant mutant mice without the marker cassette (+/m) were backcrossed into the C57BL/6J strain. c Southern blot and PCR analysis showing the insertion (+/PB-Neo) and excision (+/m and +/m*) of the piggyBac transposon. *, a reintegrated PB-Neo fragment. d Genomic DNA sequences of WT (+/+) and Per2 E′-box mutant (+/m and m/m) mice, illustrating the precise mutation of the E′-box and seamless excision of the PB-Neo cassette from the TTAA site. Heterozygous (+/m) mouse sequences exhibit dual signals for CACGTT and GCTAGT at the E′-box. e ChIP values for WT (+/+) and homozygous mutant (m/m) mouse liver sampled at 4 h intervals for 24 h on the first day in DD. Values are means ± s.e.m. of three technical replicates. Source data for (c) and (e) are provided as a Source Data file

a b Hours after synchronization with DEX 1.4 Per2 Intron1.2 Per2 1.2 1.0 +/+ 0 4 8 12 16 20 24 28 32 36 4044 48 52 56 60 64 68 72 76 80 1.0 0.8 +/+ 250 0.8 m/m 0.6 0.6 0.4 0.4 PER2 0.2 0.2 0 0 150 48 56 64 72 80 88 48 56 64 72 80 88 β-Actin 42 1.4 Per11.2 Per31.2 Cry1 1.2 1.0 1.0 0 4 8 121620242832364044 48 52 56606468 72 76 80 1.0 0.8 0.8 m/m 0.8 250 0.6 0.6 0.6 0.4 0.4 0.4 0.2 0.2 0.2 PER2 0 0 0 150 48 56 64 72 80 88 48 56 64 72 80 88 48 56 64 72 80 88 β 42 -Actin 1.4 Cry21.4 Clock1.4 Bmal1 kDa 1.2 1.2 1.2 1.0 1.0 1.0 0.8 0.8 0.8 1 0.6 0.6 0.6 0.4 0.4 0.4 0.8 +/+ m/m Relative mRNA levels 0.2 0.2 0.2 0 0 0 0.6 48 56 64 72 80 88 48 56 64 72 80 88 48 56 64 72 80 88 1.2 Nr1d11.2 Dbp1.4 E4bp4 0.4 1.0 1.0 1.2 0.8 0.8 1.0 0.8

-Actin normalized) 0.6 0.6 Relative intensity 0.2 0.6 β

( 0.4 0.4 0.4 0 0.2 0.2 0.2 0 0 0 0 4 8 12 16 20 24 28 32 36 4044 48 52 56 60 64 68 72 76 80 48 56 64 72 80 88 48 56 64 72 80 88 48 56 64 72 80 88 Hours after DEX Hours after DEXHours after DEX Hours after DEX

Fig. 2 Per2 E′-box is essential for maintaining cell-autonomous circadian oscillations. a Temporal profiles of PER2 protein expression in Per2E′+/+ and Per2E ′m/m fibroblasts. Representative immunoblots and normalized densitometry values (n = 3, mean ± s.e.m.) are shown. b mRNA profiling of clock genes in Per2E′+/+ and Per2E′m/m fibroblasts (n = 2, for each data point). For Per2, both intron and exon RNA were analyzed. The data are presented as the mean ± variation. Source data are provided as a Source Data file In the Per2E′m/m cells, we observed that both Per2 mRNA to Per2 transcription. The dysfunctional Per2 E′-box also and pre-mRNA (intronic RNA) levels remained constitutively abolished the circadian expression of other clock genes and high, with values exceeding those of the circadian peak in the output genes, including Per1, Per3, Cry1, Cry2, Bmal1, Nr1d1, WT cells at 72 h (Fig. 2b). Thus, Per2 transcription remains Dbp,andE4bp4 (Fig. 2b). These pervasive effects of the active even without this particular E′-box sequence in the mutation provide evidence that the Per2 E′-box is a funda- promoter. Moreover, extensive mRNA profiling revealed that mental cis-element that maintains normal molecular clock the effects of the mutation of the Per2 E′-box were not limited oscillations.

NATURE COMMUNICATIONS | (2019) 10:2563 | https://doi.org/10.1038/s41467-019-10532-2 | www.nature.com/naturecommunications 3 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-10532-2

a SCN Per2 E′ +/+ (WT) Per2 E′ m/m Per2 E′ m/m ; Per1+/– Per1+/– 5 5 5 5 cpm)

4 0 0 0 0 Bmal1-Eluc (1×10 luminescence –5 –5 –5 –5 012345678 012345678 012345678 012345678 Days in culture Days in culture Days in culture Days in culture

b c 7 6 0.8 ** ** 5

cpm) 4 0.6 4 * 3

0.4 variation 2 FFT Max-to-min (1×10 1

amplitude 0.2 0 0 Day 1 2 3 4561 2 34561 2 34561 2 3456 Per1 +/+ +/++/− +/− Per1 +/+ +/+ +/− +/− Per2 E′ +/+ m/m m/m +/+ Per2 E′ +/+m/m m/m +/+ d Lung Per2 E′ +/+ (WT) Per2 E′ m/m Per2 E′ m/m ; Per1+/– Per1+/– 3 3 3 3 cpm) 4 0 0 0 0 Bmal1-Eluc (1×10 luminescence

–3 –3 –3 –3 012345678 012345678 012345678 012345678 Days in culture Days in culture Days in culture Days in culture

e f 6 5 0.8 ** ** 4

0.6 cpm) 3 4 * 2

0.4 variation FFT Max-to-min

(1×10 1

amplitude 0.2 0 0 Day145632 3 12 456145632 3 14562 Per1 +/+ +/+ +/− +/– Per1 +/+ +/++/− +/− Per2 E′ +/+ m/m m/m +/+ Per2 E′ +/+m/m m/m +/+

Fig. 3 Per2 E′-box is essential for sustainable Bmal1 oscillations in SCN and lung explants. a Bmal1-Eluc bioluminescence traces of ex vivo SCN cultures from Per2E′+/+ (n = 5), Per2E′m/m (n = 5), Per2E′m/m; Per1+/− (n = 5), and Per1+/− (n = 3) mice. Averaged de-trended data are shown. b FFT amplitude of (a). *P < 0.05, **P < 0.001, one-way ANOVA, Bonferroni post hoc test. c Daily max-to-min variations of (a). The data are presented as the mean ± s.e.m. d Bmal1-Eluc bioluminescence traces of lung tissue explant cultures from Per2E′+/+ (n = 5), Per2E′m/m (n = 5), Per2E′m/m; Per1+/− (n = 5), and Per1+/− (n = 3) mice. Averaged de-trended data are shown. e FFT amplitude of (d). *P < 0.05, **P < 0.001, one-way ANOVA, Bonferroni post hoc test. f Daily max- to-min variations of (d). The data are presented as the mean ± s.e.m. Source data for (c) and (f) are provided as a Source Data ﬁle

Per2 E′-box is essential to maintain molecular oscillations in damped within 2–3 cycles (Fig. 3a). Fast Fourier transform (FFT) the SCN. The SCN in the hypothalamus is the primary regulator analysis of the de-trended waveforms of days 1.5−7.5 (Fig. 3b) of daily rhythms of behavior and physiology in mammals24,25. confirmed the reduced rhythmicity in the Per2E′m/m SCN. This We next examined the effect of the Per2 E′-box mutation on the finding indicated the relevance of the Per2 E′-box in maintaining SCN clock. To study the molecular rhythms in the SCN, we used normal molecular circadian oscillations in the SCN. Nevertheless, a Bmal1 promoter-driven luciferase reporter (Bmal1-Eluc)26,asit detectable rhythms were noted in the attenuated Per2E′m/m SCN allowed the assessment of molecular rhythms that are not a slices over the initial 2–3 cycles (Fig. 3a). These remaining simple reflection of the Per2 loop. In agreement with previous rhythms were likely dependent on Per1, because SCN explants reports27, organotypic SCN slices prepared from control mice from Per2E′m/m; Per1 heterozygous null mice (Per2E′m/m; (Per2E′+/+; Bmal1-Eluc) displayed persistent circadian rhythms Per1+/−;Bmal1-Eluc) exhibited even fewer persistent rhythms, of luminescence, which continued for over a week in culture which were damped within 2 cycles (Fig. 3a–c). Importantly, (Fig. 3a, see also Supplementary Fig. 4). In contrast, all tested similar attenuation of Bmal1-Eluc rhythmicity by Per2E′m/m and SCN slices from the Per2 E′-box mutant mice (Per2E′m/m;Bmal1- further by Per1+/− could be reproduced in cultures of lung Eluc) displayed attenuated rhythms of luminescence, which were (Fig. 3d–f) and adrenal explants (Supplementary Fig. 4),

4 NATURE COMMUNICATIONS | (2019) 10:2563 | https://doi.org/10.1038/s41467-019-10532-2 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-10532-2 ARTICLE

a +/+ m/m b Body temperature +/+ Time (h) m/m Time (h) +/+ 0.8 ** 012012 0 0 12 0 12 0 0 12 0 12 0 CT12 8 ** 10 –4 –4 0.6 12 618 –2 –2 +/+ 14 0 0 16 2 2 18 0

0.4 Days in LL 4 4 20 Vector* 6 6 012012 0 8 8 length

10 10 FFT power m/m Days 12 12 0.2 8 14 14 10 CT12 16 16 12 18 18 m/m 14 20 20 0 16 618 18 Days in LL 1–15 4–18 7–21 Days in LL 20 0

c PS (days) +/+ Time (h) m/m Time (h) Time (h) 50 0 12 0 12 0 0 12 0 12 0 2 4 6 8 10 12 14 08246

–6 –6 –2 –4 –4 0 +/+ –2 –2 2 0 0 2 2 4

4 4 Days 6 m/m Days 6 6 +/+ *** 8 8 8 10 10 10 m/m 12 12

d +/+ Time (h) m/m Time (h) Time (h) 0 12 0 12 0 0 12 0 12 0 20 01248 +/+ m/m –4 –4 –2 –2 –2 0 0 * 0 * 2 *** 2 2 4 4 4 Days 6 Days 6 6 +/+ 8 6 4 2 0 10 8 8 8 m/m 12 –2 10 10 10 Phase shift (h) 12 12

Fig. 4 a Destabilized circadian locomotor activity and body temperature rhythms of Per2 E′-box mutant mice. Representative locomotor activity records of Per2E′+/+ and Per2E′m/m mice under light/dark (LD) followed by constant light (LL). The graph shows changes in FFT power in LL. **P < 0.01, two-way repeated-measures ANOVA, Bonferroni post hoc test (Per2E′+/+, n = 7; Per2E′m/m, n = 9). b Representative body temperature records of Per2E′+/+ and Per2E′m/m mice in LL. The mean temperature of the entire data series was calculated, and the data points above the mean are plotted. Rayleigh plots show phase distribution of elevated body temperature on days 7–21 (n = 3, each genotype). Data from independent animals are color-coded. Arrow length reflects the r-value of each distribution. *P < 0.05, two-tailed unpaired t-test. c Representative records of locomotor activity (left) and plots of activity onset +/+ m/m (middle) of Per2E′ and Per2E′ mice before and after an 8-h phase advance in LD cycles. The PS50 values represent the time required for 50% phase-shift (right). ***P < 0.0001, two-tailed unpaired t-test (Per2E′+/+, n = 7; Per2E′m/m, n = 9). d Representative locomotor activity records (left), activity onset (middle), and magnitude of phase-shift (right) of Per2E′+/+ and Per2E′m/m mice subjected to an 8-h phase advance on day 1 and released to DD. ***P < 0.0001, two-tailed unpaired t-test (Per2E′+/+, n = 7; Per2E′m/m, n = 11). The data are the mean ± s.e.m. Source data are provided as a Source Data file confirming the importance of the Per2 E′-box in keeping sus- Compared to the modest phenotype in DD, we observed tained molecular circadian oscillations not only for the SCN but unstable rhythmicity of the mutant mice in LL (Fig. 4a). When also for the peripheral tissues. placed in LL for 3 weeks, WT subjects (n = 7) stayed rhythmic with a period longer than 24 h, which is consistent with data from previous reports30,31. However, none of the Per2E′m/m mice Destabilized organismal rhythms of animal lacking Per2 (n = 9) remained rhythmic (Fig. 4a, see also Supplementary E′-box. Finally, to assess the effects of the mutation at the Fig. 7); all tested mutant mice displayed behaviors characterized behavioral level, C57BL/6J-backcrossed WT and Per2E′m/m mice by a gradual decrease in the power of rhythmicity (FFT were housed in a 12-h light:12-h dark (LD) cycle and then sub- spectrogram, Fig. 4a). Since light has a negative masking effect jected to DD, LL, or experimental jet-lag conditions. Actograms on locomotor activity, we measured core body temperature of the animals in the respective conditions were acquired (see fluctuations and confirmed that the body temperature recapitu- Supplementary Figs. 5–9). Interestingly, mutant mice kept in DD lated the reduced rhythmicity of the mutant mice in showed overt circadian locomotor activity rhythms, although LL conditions (Fig. 4b, Rayleigh vector length of WT vs. with a significantly shortened free-running circadian period Per2E′m/m: 0.64 ± 0.06 vs. 0.42 ± 0.02). relative to WT mice (WT vs. Per2E′m/m: 23.71 ± 0.02 h vs. 23.50 The Per2 E′-box mutant mice were also distinct under ± 0.02 h; see Supplementary Fig. 5). This contrast with the in vitro experimental jet-lag conditions (Fig. 4c, see also Supplementary data suggests that a compensatory mechanism functions in vivo. Fig. 8). When the ambient LD cycle was advanced by 8 h, WT We thus examined the circadian behavior of Per1−/−; Per2E′m/m mice re-entrained progressively over 8 to 9 days32. In contrast, the double-deficient mice (see Supplementary Fig. 6). Although their Per2E′m/m mice adapted to the new cycle within 2–3 days. The 32 period length was unstable and changed gradually over extended 50% phase-shift value (PS50) , measured from activity onset, exposure to constant darkness, Per1−/−; Per2E′m/m mice dis- indicated a rapidity of 5.33 ± 0.28 days for WT mice and 1.26 ± played behavioral rhythms over 40 days. With analogy, Droso- 0.17 days for Per2E′m/m mice (Fig. 4c). Notably, the difference phila strains expressing period or timeless under a constitutive between the two genotypes was even pronounced when the promoter were reported to show behavioral rhythms with an animals were exposed to a single 8-h light advance, followed by altered period length12,28. These data suggest that organisms DD conditions (Fig. 4d, see also Supplementary Fig. 9). Under might have a compensatory mechanism to mitigate defective this light regime, all tested mutant mice shifted forward by transcriptional feedback regulation4,24,29. approximately 12 h, which is nearly 180° out of phase in a 24-h

NATURE COMMUNICATIONS | (2019) 10:2563 | https://doi.org/10.1038/s41467-019-10532-2 | www.nature.com/naturecommunications 5 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-10532-2 cycle, relative to WT mice (Fig. 4d). The underlying mechanism these near-native conditions, we noticed that the deletion of the Per2 of this extremely large shift is unknown. While this could be due E′-box leads to accumulation of Per2 mRNA and protein in cultured to an indirect effect of the weak clock, a limit cycle oscillator fibroblasts. Constitutively un-suppressed transcription of Per2 likely model33 predicts that a reduced-amplitude pacemaker in the underlies the constant accumulation of PER2 protein and resultant mutant mice (reduced radius of the limit cycle) could have this compromised gene expression rhythms in the mutant cells. effect34. We also noticed that after a light pulse exposure, the Cis-element provides a place where both active and repressive mutant mice show a slightly enhanced Per2 mRNA induction in transcription complexes are recruited. The net effect of its the SCN (see Supplementary Fig. 10). This perhaps also partly absence thus seems context-dependent. The question of this contributes to the enhanced resetting of the mutant mice. context specificity is still unresolved in the field of transcription and chromatin remodeling. Particularly, in our case, the mechanism(s) that allows the continued transcription of Per2 Discussion without relying on the E′-box is still unknown. It is possible that Non-coding cis-regulatory elements are known to play a critical native chromatin structure of the Per2 might permit its tran- role in development, but their precise effects on daily behavior scription. It is also conceivable that other circadian cis-elements and physiology in adulthood remain elusive. The present study on the Per2 promoter, such as D-box and CRE, might contribute was designed to provide genetic evidence for the roles of the to the basal transcription of Per29,18,42. In this regard, the con- circadian cis-element in vivo. Despite the presence of compen- tinued expression of Per2 might reflect a confounding effect of satory mechanisms in vivo, our work shows that the Per2 E′-box being chronically deficient in the functional E′-box. Given that is essential for maintaining optimal locomotor activity rhythms Per2 is under regulation of interlocked feedback cycles6,9, under LL conditions and enabling the phase of the clock to resist remaining Per2 transcription might be a homeostatic con- abrupt shifts in LD cycling. In general, adapting to new LD cycles sequence of a nonfunctional clock in the mutant cells. A complete is rarely instantaneous and requires repeated 24 h cycles. How- understanding of circadian regulation of Per2 in vivo under ever, the Per2 E′-box mutant mice adapted to a new phase native conditions remains a challenge for future study. immediately. These data elucidate the impact of the deletion of Genome-wide association studies recently identified many non- the non-coding cis-element in daily maintenance of behavioral coding variants that account for human chronotypes43–45.Giventhe activity in adulthood. physiologic relevance of the Per2 cis-element, a targeted point In addition, our cell culture data reinforce the concept that mutation strategy would facilitate hypothesis-driven approaches to circadian clock phenotypes are more drastic at the cellular level understand the extent of the impact of the non-coding elements on than the organismic level4,24. Per2 E′-box mutation leads to the daily physiology and pathophysiology of the organism. unsustainable gene expression rhythms in organotypic SCN slices and cultured peripheral tissues and fibroblasts. These observa- Methods tions substantiate the general conjecture that cis-regulatory Generation of the Per2 E′-box mutant mice. The CACGTT E′-box sequence element-based gene transcription is essential for sustaining cel- located in the Per2 promoter (–20 to –15; +1, the putative transcription start site17) ′ lular clock oscillations. was targeted. Note that numbering of the relative position of the E -box sequence will change according to the definition of the transcription initiation site17,18 (see Not surprisingly, the overall behavioral phenotypes of the Per2 also RefSeq database in the UCSC genome browser: https://genome.ucsc.edu/). E′-box mutant mice reveal that the Per2 E′-box is not an absolute CACGTT was mutated to GCTAGT, because this change is reported to abrogate requirement for behavioral rhythm generation. Under DD con- CLOCK:BMAL1-mediated activation of Per2 promoter activity in vitro17.We 46,47 ditions, its deletion affects only the circadian period (Supple- employed piggyBac (PB) transposon-based gene engineering system , because conventional methods relying on Cre/loxP or Flp/FRT systems leave behind a mentary Figs. 5 and 6), a phenotype analogous to that observed in single loxPorFRT-derived sequence after excision, which causes the creation of transgenic Drosophila strains expressing period or timeless via a ‘footprint’ (residual ectopic sequence) that might affect promoter architecture and/ constitutive promoter12,28. At the gene expression level, Per2 or enhancer communication of target gene48. The targeting vector to generate PB expression in the liver and SCN in the DD-kept mutant mice mutant allele of Per2 was constructed by using a Red/ET recombination system (Gene Bridges). The bacterial artificial chromosome (BAC) containing Per2 was in vivo was still rhythmic, albeit with an increased baseline (see obtained from BACPAC Resources at the Children’s Hospital and Research Center Supplementary Fig. 11). These observations suggest that organ- at Oakland (RP23-343F13). With a Red/ET cloning method (Gene Bridges), a 10- isms have a compensatory mechanism to mitigate defective kb genomic region of Per2 (–5899 to +4,103; +1, the putative transcription start site17) that corresponds to the region of 6.5 kb upstream (long arm) and 3.5 kb transcriptional feedback regulation. Correspondingly, previous ′ – – studies demonstrate that behavioral rhythms do not necessarily downstream (short arm) of the Per2 E -box ( 20/ 15) was cloned into a vector fl 29,35–37 containing a diphtheria toxin A gene (pDTA vector). Then, a DNA amplicon of re ect cellular clock phenotypes ; superior circadian per- Per2 (−90 to +650) that contains a mutated E′-box (GCTAGT) at −20/−15 was formance of behavioral rhythmicity has been observed in several obtained by fusion PCR, and this was further modified by inserting a PB terminal clock gene mutant mice, compared to tissue cultures obtained repeat sequence (5′TR and 3′TR)-flanked neomycin cassette (PB-NEO) into the + + from the same animal strains29. In vivo multicellular and/or TTAA quadruplet sequence of Per2 ( 600/ 603). With a second-round Red/ET cloning, this fragment was recombined into the 10-kb Per2 BAC/pDTA vector. The inter-organ systemic circuitry might compensate for poor core resultant Per2 targeting vector was verified by DNA sequencing. clock function within individual cells4. In this regard, it is worth Gene targeting was carried out with TT2 embryonic stem cell. Germline noting that systemic extracellular signals are known to affect the transmission was verified by PCR and Southern blotting. The Per2 E′-box mut PB- ′ NEO F1 mice (Accession No. CDB1060K: http://www2.clst.riken.jp/arg/mutant% Per2 promoter via non-E -box cis-regulatory elements, such as 49 2+ 38 20mice%20list.html) were intercrossed with ROSA26-PBase knock-in mice (a gift Ca /cAMP response element (CRE) and glucocorticoid from A. Bradley, Wellcome Trust Sanger Institute, UK) to remove the PB-NEO 39 response element (GRE) . Extracellular circadian feedback cassette from the genome. A seamless excision of the cassette was verified by direct pathways through these non-E′-box elements might contribute to DNA sequencing. Genotypes of the E′-box were determined by DNA sequencing the compensation mechanisms in vivo40,41. and/or TaqMan qPCR with the following probes: WT, 5′-FAM-TAG TGG AAA ACG TGA CCG C-MGB-3′, and mutant, 5′-VIC-TAG TGG AAA CTA GCA CCG Our work differs from transgenic studies in which non-native C-MGB-3′. The use of different reporter dyes with separated emission wavelength 12,17 promoters are used . Transgenic studies and rescue experiments maxima (FAM and VIC) enabled concurrent detection of the two alleles in a single can be affected by shorter promoter regulatory sequences and PCR with a common primer set for Per2: forward, 5′-GGA GCC GCT AGT CCC position effects. In comparison, native promoters are characterized AGT AG-3′; reverse, 5′-AGG TGG CAC TCC GAC CAA T-3′. The established mutant mice were backcrossed into the C57BL/6J background using a marker- by endogenous enhancer elements and normal chromatin structure, assisted breeding (i.e., speed congenic) approach50,51. Where specified, Per2E′m/m allowing for the preservation of regulation by epigenetic factors, mice were intercrossed with mice carrying a Bmal1-Eluc reporter26,27,52 and/or which are known to be crucial for controlling transcription5.Under Per1-deficient mice15 of C57BL/6J background (The Jackson Laboratory, 10491).

6 NATURE COMMUNICATIONS | (2019) 10:2563 | https://doi.org/10.1038/s41467-019-10532-2 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-10532-2 ARTICLE

All animal experiments were conducted in compliance with ethical regulations in TG GAC TTC CAG GTC G-3′. for Gapdh, probe, FAM-CGG CCA AAT CCG Kyoto University and performed under protocols approved by the Animal Care TTC ACA CCG ACC-TAMRA, fw: 5′-GAG ACG GCC GCA TCT TCT T-3′, rv: and Experimentation Committee of Kyoto University and Institutional Animal 5′-TCT CCA CTT TGC CAC TGC A-3′. Values were normalized to average Care and Use Committee of RIKEN Kobe Branch. expression of Rplp0 and Gapdh.

ChIP. ChIP assays were performed with technical replicates19,53: We repeated Real-time bioluminescence recording and data analysis. Tissues from mice 26 chromatin/DNA shearing with equal amounts of aliquots from the same liver carrying a Bmal1 promoter-driven luciferase reporter (Bmal1-ELuc) were har- – – nuclear sample. In brief, whole liver sample was homogenized with ice-cold PBS (4 vested from adult (2 3 months-old) male mice between ZT10 12 and cultured 58 ml/liver) and cross-linked in two steps using first 2 mM disuccinimidyl glutarate according to a published method . In brief, brains were sliced coronally into 400- μ for 20 min then 1% methanol-free ultrapure formaldehyde for 5 min at room m sections, and the paired SCN, which contains minimal volumes of the non- 59 3 59 temperature. Glycine was added for 5 min in 125 mM final concentration. The SCN, were cultured . The lungs were cut into a small piece (~2 × 2 × 0.3 mm ) . homogenates (~5 ml/liver) were mixed with ice-cold 2.3 M sucrose buffer (10 ml/ The adrenal glands were sliced into a section of 0.3 mm thickness. All tissues were liver) containing 125 mM glycine, 10 mM HEPES pH 7.6, 15 mM KCl, 2 mM cultured separately on a Millicell membrane (PICMORG50, Millipore) with EDTA, 0.15 mM spermine, 0.5 mM spermidine, 0.5 mM DTT, and 0.5 mM PMSF DMEM medium containing10 mM HEPES (pH 7.2), 2% B27 (Invitrogen) and 1 and layered on top of a 5 ml cushion of 1.85 M sucrose. Ultracentrifugation was mM luciferin, in 35-mm dish, and air sealed. Bioluminescence recording was performed at 105,000 × g for 1 h at 4 °C with a Beckman SW28 rotor. The resultant started immediately upon placement in culture with a dish-type luminometer nuclear pellets were stored at −80 °C until use. The nuclei were resuspended in 1.5 (Kronos Dio, ATTO) maintained at 37 °C. Luminescence was measured for 3 min ml per liver of IP buffer (10 mM Tris-HCl pH 7.5, 150 mM NaCl, 1 mM EDTA, 1% for each dish at 30-min intervals. The data were smoothed using a 2 h moving Triton X-100, 0.1% sodium deoxycholate, 1 mM PMSF, protease inhibitor cocktail) average and further detrended by subtracting a 24 h running average for FFT – and divided equally into three aliquots, which were each separately sonicated analysis. A fourth-order Blackman Harris window was applied before the power around 15 s for 80 times at 4 °C using a Bioruptor UCW-201TM apparatus (Tosho spectrum calculation. The spectrum was normalized to an integral of one by Denki, Japan). For each reaction, 10 μg fragmented chromatin was resuspended in dividing each of its elements by the sum of all elements. Circadian rhythmicity was fi – 500 μl of IP buffer, pre-cleared with protein A-agarose, and incubated overnight at de ned as relative spectral power density at the peak in the circadian range (20 30 4 °C with the following antibody: for PER1, 2 μl of anti-mPER1 rabbit antiserum h). The peak values represent the power within a frequency band of 0.009. (Millipore, #AB2201); for CRY1, 1 μgofaffinity-purified anti-mCRY1 guinea pig polyclonal antibody19,54; for CRY2, 1 μgofaffinity-purified anti-mCRY2 guinea pig Locomotor activity recording and data analysis. C57BL/6J-backcrossed Per2E′m/+ polyclonal antibody19,54; and for CLOCK, 1 μg of anti-mCLOCK mouse mono- mice were intercrossed to produce homozygous mutant (Per2E′m/m) and WT clonal antibody55 (MBL, #D349-3). To collect immunocomplexes, 40 μl of Protein (Per2E′+/+) progenies using in vitro fertilization. We used adult male mice (8- to A/G Plus-agarose beads (Santa Cruz) were added and further incubated at 4 °C for 10-week old). The animals were housed individually in light-tight, ventilated closets 1.5 h. After beads washing and DNA elution53, eluted DNA fragments were pur- under indicated lighting conditions with ad libitum access to food and water. ified with Qiaquick Nucleotide Removal Kit (Qiagen) and subjected to qPCR Locomotor activity was recorded via passive infrared sensors (PIRs, FA-05F5B; analysis with Per2 E′-box primers (forward primer, 5ʹ-GAG GTG GCA CTC CGA Omron) with 1-min resolution and analyzed with CLOCKLAB software (Acti- CCA ATG-3ʹ; reverse primer, 5ʹ-GCG TCG CCC TCC GCT GTC AC-3ʹ) and Per2 metrics)60. Free-running period in DD was determined with χ2 periodogram, based −4.5 kb negative binding site primers (forward primer, 5ʹ-CCA CAC GGT ACT on animal behaviors in a 40-day interval taken 3 days after the start of DD con- CAG CGG GC-3ʹ; reverse primer, 5ʹ-GGG TCA CTG CGA GCC TTG CC-3ʹ). dition. Rhythmicity in LL was evaluated using FFT-Relative Power (CLOCKLAB, Actimetrics). A fourth-order Blackman–Harris window was applied before the power spectrum calculation. The spectrum was normalized to an integral of one by ′m/m fi Cell culture and immunoblotting. WT and Per2E primary broblasts were dividing each of its elements by the sum of all elements. Circadian rhythmicity was isolated from the lung tissue of adult male mice according to a protocol established defined as relative spectral power density at the peak in the circadian range (20–36 by Seluanov et al.56. Cells used for time course assay were between passages 4 and 5 h). The peak values represent the power within a frequency band of 0.006. Speed of 6. Dispersed cells were uniformly plated in 24-well plates at a density of 1 × 10 32,61 behavioral re-entrainment was evaluated using 50% phase-shift value (PS50) . cells per well and cultured for 2 days prior to synchronization, for which cells were = fi To determine PS50, sigmoidal dose-response curve with variable slope, Y Bottom treated with dexamethasone (DEX, nal concentration, 200 nM) for 3 h, followed + (Top – Bottom)/(1 + 10(log PS50 – X) HillSlope), was fitted to the onset time points by medium refreshment at Time0. Cells were harvested every 4 h in either TRIzol using GraphPad Prism software. reagent (Invitrogen) for RNA analysis or 2 × Laemmli buffer for Western blot analysis. Immunoblotting was performed using an affinity-purified anti-mPER2 rabbit polyclonal antibody23 (final concentration, 2 μgml−1) and β-Actin antibody Body temperature recording and data analysis. Precalibrated temperature data (A5441, Sigma, 1:1000). loggers (Thermochron iButtons, DS1921H, Maxim) were surgically implanted to the peritoneal cavity of mice under general anaesthesia. After a week of recovery in LD, mice were transferred into LL. The iButtons were programmed for collecting RNA extraction and RT-qPCR. Total RNA was extracted with RNeasy kit (Qia- temperature data every 20 min. For data analysis, the mean temperature of the gen) and converted to cDNA with SuperScript VILO cDNA Synthesis kit (Invi- entire data series during days 7 to 21 in LL was calculated, and the data points trogen). qPCR was run on a BioMark HD System (Fluidigm) with a 48.48 Fluidigm above the mean were displayed in double-plotted format using CLOCKLAB. For BioMark Dynamic Array chip (Fluidigm)57. TaqMan probe and primer sets were as circular plot, data were detrended by subtracting a 24 h running average, and follows; for Per2, probe, FAM-AGG CAC CTC CAA CAT GCA ACG AGC C- periods of top 20% temperature were deployed in Rayleigh format using Oriana TAMRA, forward primer (fw): 5′-GCA CAT CTG GCA CAT CTC GG-3′, reverse 4 software (Kovacs Computer Services, UK). CT12 was extrapolated from the primer (rv): 5′-TGG CAT CAC TGT TCT GAG TGT C-3′; for Per2 intron, probe, onsets of locomotor activity during days 3–6. Phase distribution was assessed by FAM-TGG AGC CCA CCG CAG ACA GCC C-TAMRA, fw: 5′-CCT CTC ACC mean vector length (Oriana 4). TCA TGC CCT TTT AG-3′, rv: 5′-CTG CTC AGA CCA ACA GAT TTA TCA-3′; for Per1, probe, FAM-AGC CCC TGG CTG CCA TGG-TAMRA, fw: 5′-CAG GCT Reporting summary. Further information on research design is available in TCG TGG ACT TGA GC-3′, rv: 5′-AGT GGT GTC GGC GAC CAG-3′; for Per3, the Nature Research Reporting Summary linked to this article. probe, FAM-TTC TGC TCA TCA CCA CCC TGC GGT TCC-TA MRA, fw: 5′-ACA GCT CTA CAT CGA GTC CAT G-3′, rv: 5′-CAG TGT CTG AGA GGA AGA AAA GTC-3′; for Cry1, probe, FAM-TGA TCC ACA GGT CAC Data availability CAC GAG TCA GGA A-TAMRA, fw: 5′-TAG CCA GAC ACG CGG TTG-3′, rv: The source data underlying Figs. 1c, e, 2a, b, 3c, f, 4a–d, Supplementary Figs. 3, 4e, 6b, c, 5′-AGC AGT AAC TCT TCA AAG ACC TTC A-3′; for Cry2, probe, FAM-AGG 10a, b, 11a, b are provided as a Source Data file. All the other data supporting the findings ′ TCT CTC ATA GTT GGC AAC CCA GGC-TAMRA, fw: 5 -TGG ACA AGC ACT of this study are available within the article and its Supplementary Information files and ′ ′ ′ TGG AAC GG-3 , rv: 5 -GGC CAG TAA GGA ATT GGC ATT C-3 ; for Clock, from the corresponding authors upon reasonable request. A reporting summary for this ′ probe, FAM-ACC CAG AAT CTT GGC TTT TGT CAG CAG C-TAMRA, fw: 5 - article is available as a Supplementary Information file. TGG CAT TGA AGA GTC TCT TCC TG-3′, rv: 5′-GAG ACT CAC TGT GTT GAT ACG ATT G-3′; for Bmal1, probe, FAM-CGC CAA AAT AGC TGT CGC CCT CTG ATC T-TAMRA, fw: 5′-GTA CGT TTC TCG ACA CGC AAT AG-3′, Received: 6 October 2018 Accepted: 16 May 2019 rv: 5′-GTA CCT AGA AGT TCC TGT GGT AGA-3′; for Nr1d1, probe, FAM-CCC TGG ACT CCA ATA ACA ACA CAG GTG G-TAMRA, fw: 5′-TCA GCT GGT GAA GAC ATG ACG-3′, rv: 5′-GAG GAG CCA CTA GAG CCA ATG-3′; for Dbp, probe, FAM-CGG CTC CCA GAG TGG CCC GC-TAMRA, fw: 5′-CGG CTC TTG CAG CTC CTC-3′, rv: 5′-GTG TCC CTA GAT GTC AAG CCT G-3′; for E4bp4, probe, FAM-CAG GGA GCA GAA CCA CGA TAA CCC ATG A-TAMRA, fw: 5′-CGC CAG CCC GGT TAC AG-3′, rv: 5′-CAT CCA TCA ATG References GGT CCT TCT G-3′; for Rplp0, probe, FAM-TGG CAA TCC CTG ACG CAC C 1. Kvon, E. Z. et al. Progressive loss of function in a limb enhancer during snake GC C-TAMRA, fw: 5′-GCG TCC TCG TTG GAG TGA C-3′, rv: 5′-AAG TAG T evolution. Cell 167, 633–642 e611 (2016).

NATURE COMMUNICATIONS | (2019) 10:2563 | https://doi.org/10.1038/s41467-019-10532-2 | www.nature.com/naturecommunications 7 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-10532-2

2. Indjeian, V. B. et al. Evolving new skeletal traits by cis-regulatory changes in 34. Vitaterna, M. H. et al. The mouse Clock mutation reduces circadian bone morphogenetic proteins. Cell 164,45–56 (2016). pacemaker amplitude and enhances efficacy of resetting stimuli and phase- 3. Whyte, W. A. et al. Master transcription factors and mediator establish super- response curve amplitude. Proc. Natl Acad. Sci. USA 103, 9327–9332 (2006). enhancers at key cell identity genes. Cell 153, 307–319 (2013). 35. Liu, C., Weaver, D. R., Strogatz, S. H. & Reppert, S. M. Cellular construction of 4. Rosbash, M. et al. Transcriptional feedback and definition of the circadian a circadian clock: period determination in the suprachiasmatic nuclei. Cell 91, pacemaker in Drosophila and animals. Cold Spring Harb. Symp. Quant. Biol. 855–860 (1997). 72,75–83 (2007). 36. Herzog, E. D., Takahashi, J. S. & Block, G. D. Clock controls circadian period in 5. Masri, S. & Sassone-Corsi, P. The circadian clock: a framework linking isolated suprachiasmatic nucleus neurons. Nat. Neurosci. 1,708–713 (1998). metabolism, epigenetics and neuronal function. Nat. Rev. Neurosci. 14,69–75 37. Nakamura, W., Honma, S., Shirakawa, T. & Honma, K. Clock mutation (2013). lengthens the circadian period without damping rhythms in individual SCN 6. Takahashi, J. S. Transcriptional architecture of the mammalian circadian neurons. Nat. Neurosci. 5, 399–400 (2002). clock. Nat. Rev. Genet. 18, 164–179 (2017). 38. Travnickova-Bendova, Z., Cermakian, N., Reppert, S. M. & Sassone-Corsi, P. 7. Hao, H., Allen, D. L. & Hardin, P. E. A circadian enhancer mediates PER- Bimodal regulation of mPeriod promoters by CREB-dependent signaling and dependent mRNA cycling in Drosophila melanogaster. Mol. Cell. Biol. 17, CLOCK/BMAL1 activity. Proc. Natl Acad. Sci. USA 99, 7728–7733 (2002). 3687–3693 (1997). 39. So, A. Y., Bernal, T. U., Pillsbury, M. L., Yamamoto, K. R. & Feldman, B. J. 8. Gekakis, N. et al. Role of the CLOCK protein in the mammalian circadian Glucocorticoid regulation of the circadian clock modulates glucose mechanism. Science 280, 1564–1569 (1998). homeostasis. Proc. Natl Acad. Sci. USA 106, 17582–17587 (2009). 9. Ueda, H. R. et al. System-level identification of transcriptional circuits 40. Hastings, M. H., Maywood, E. S. & O’Neill, J. S. Cellular circadian pacemaking underlying mammalian circadian clocks. Nat. Genet. 37, 187–192 (2005). and the role of cytosolic rhythms. Curr. Biol. 18, R805–R815 (2008). 10. O’Neill, J. S. & Reddy, A. B. Circadian clocks in human red blood cells. Nature 41. Parsons, M. J. et al. The regulatory factor ZFHX3 Modifies Circadian function 469, 498–503 (2011). in SCN via an AT motif-driven axis. Cell 162, 607–621 (2015). 11. Edgar, R. S. et al. Peroxiredoxins are conserved markers of circadian rhythms. 42. O’Neill, J. S., Maywood, E. S., Chesham, J. E., Takahashi, J. S. & Hastings, M. Nature 485, 459–464 (2012). H. cAMP-dependent signaling as a core component of the mammalian 12. Yang, Z. & Sehgal, A. Role of molecular oscillations in generating behavioral circadian pacemaker. Science 320, 949–953 (2008). rhythms in Drosophila. Neuron 29, 453–467 (2001). 43. Lane, J. M. et al. Genome-wide association analysis identifies novel loci for 13. Zheng, B. et al. The mPer2 gene encodes a functional component of the chronotype in 100,420 individuals from the UK Biobank. Nat. Commun. 7, mammalian circadian clock. Nature 400, 169–173 (1999). 10889 (2016). 14. Zheng, B. et al. Nonredundant roles of the mPer1 and mPer2 genes in the 44. Hu, Y. et al. GWAS of 89,283 individuals identifies genetic variants associated mammalian circadian clock. Cell 105, 683–694 (2001). with self-reporting of being a morning person. Nat. Commun. 7, 10448 (2016). 15. Bae, K. et al. Differential functions of mPer1, mPer2, and mPer3 in the SCN 45. Jones, S. E. et al. Genome-wide association analyses in 128,266 individuals circadian clock. Neuron 30, 525–536 (2001). identifies new morningness and sleep duration loci. PLoS Genet. 12, e1006125 16. Toh, K. L. et al. An hPer2 phosphorylation site mutation in familial advanced (2016). sleep phase syndrome. Science 291, 1040–1043 (2001). 46. Ding, S. et al. Efficient transposition of the piggyBac (PB) transposon in 17. Yoo, S. H. et al. A noncanonical E-box enhancer drives mouse Period2 mammalian cells and mice. Cell 122, 473–483 (2005). circadian oscillations in vivo. Proc. Natl Acad. Sci. USA 102, 2608–2613 47. Wang, W. et al. Chromosomal transposition of PiggyBac in mouse embryonic (2005). stem cells. Proc. Natl Acad. Sci. USA 105, 9290–9295 (2008). 18. Akashi, M., Ichise, T., Mamine, T. & Takumi, T. Molecular mechanism of cell- 48. Meier, I. D. et al. Short DNA sequences inserted for gene targeting can autonomous circadian gene expression of Period2, a crucial regulator of the accidentally interfere with off-target gene expression. FASEB J. 24, 1714–1724 mammalian circadian clock. Mol. Biol. Cell 17, 555–565 (2006). (2010). 19. Koike, N. et al. Transcriptional architecture and chromatin landscape of the 49. Rad, R. et al. PiggyBac transposon mutagenesis: a tool for cancer gene core circadian clock in mammals. Science 338, 349–354 (2012). discovery in mice. Science 330, 1104–1107 (2010). 20. Yusa, K. Seamless genome editing in human pluripotent stem cells using 50. Markel, P. et al. Theoretical and empirical issues for marker-assisted breeding custom endonuclease-based gene targeting and the piggyBac transposon. Nat. of congenic mouse strains. Nat. Genet. 17, 280–284 (1997). Protoc. 8, 2061–2078 (2013). 51. Suemizu, H. et al. Establishing EGFP congenic mice in a NOD/Shi-scid IL2Rg 21. Yusa, K., Rad, R., Takeda, J. & Bradley, A. Generation of transgene-free (null) (NOG) genetic background using a marker-assisted selection protocol induced pluripotent mouse stem cells by the piggyBac transposon. Nat. (MASP). Exp. Anim. 57, 471–477 (2008). Methods 6, 363–369 (2009). 52. Myung, J. et al. GABA-mediated repulsive coupling between circadian clock 22. Rey, G. et al. Genome-wide and phase-specific DNA-binding rhythms of neurons in the SCN encodes seasonal time. Proc. Natl Acad. Sci. USA 112, BMAL1 control circadian output functions in mouse liver. PLoS Biol. 9, E3920–E3929 (2015). e1000595 (2011). 53. Ota, T. et al. Angiotensin II triggers expression of the adrenal gland zona 23. Tainaka, M., Doi, M., Inoue, Y., Murai, I. & Okamura, H. Circadian PER2 glomerulosa-specific 3beta-hydroxysteroid dehydrogenase isoenzyme through protein oscillations do not persist in cycloheximide-treated mouse embryonic de novo protein synthesis of the orphan nuclear receptors NGFIB and fibroblasts in culture. Chronobiol. Int. 35, 132–136 (2018). NURR1. Mol. Cell. Biol. 34, 3880–3894 (2014). 24. Herzog, E. D., Hermanstyne, T., Smyllie, N. J. & Hastings, M. H. Regulating 54. Lee, C., Weaver, D. R. & Reppert, S. M. Direct association between mouse the Suprachiasmatic Nucleus (SCN) Circadian Clockwork: interplay between PERIOD and CKIepsilon is critical for a functioning circadian clock. Mol. Cell. cell-autonomous and circuit-level mechanisms. Cold Spring Harb. Perspect. Biol. 24, 584–594 (2004). Biol. 9, a027706 (2017). 55. Yoshitane, H. et al. CLOCK-controlled polyphonic regulation of circadian 25. Okamura, H. Suprachiasmatic nucleus clock time in the mammalian circadian rhythms through canonical and noncanonical E-boxes. Mol. Cell. Biol. 34, system. Cold Spring Harb. Symp. Quant. Biol. 72, 551–556 (2007). 1776–1787 (2014). 26. Noguchi, T., Ikeda, M., Ohmiya, Y. & Nakajima, Y. A dual-color luciferase 56. Seluanov, A., Vaidya, A. & Gorbunova, V. Establishing primary adult assay system reveals circadian resetting of cultured fibroblasts by co-cultured fibroblast cultures from rodents. J. Vis. Exp. 44, e2033 (2010). adrenal glands. PLoS ONE 7, e37093 (2012). 57. Fustin, J. M. et al. RNA-methylation-dependent RNA processing controls the 27. Ono, D. et al. Dissociation of Per1 and Bmal1 circadian rhythms in the speed of the circadian clock. Cell 155, 793–806 (2013). suprachiasmatic nucleus in parallel with behavioral outputs. Proc. Natl Acad. 58. Yamazaki, S. & Takahashi, J. S. Real-time luminescence reporting of circadian Sci. USA 114, E3699–E3708 (2017). gene expression in mammals. Methods Enzymol. 393, 288–301 (2005). 28. Kadener, S., Menet, J. S., Schoer, R. & Rosbash, M. Circadian transcription 59. Doi, M. et al. Gpr176 is a Gz-linked orphan G-protein-coupled receptor that contributes to core period determination in Drosophila. PLoS Biol. 6, e119 sets the pace of circadian behaviour. Nat. Commun. 7, 10583 (2016). (2008). 60. Doi, M. et al. Circadian regulation of intracellular G-protein signalling 29. Liu, A. C. et al. Intercellular coupling confers robustness against mutations in mediates intercellular synchrony and rhythmicity in the suprachiasmatic the SCN circadian clock network. Cell 129, 605–616 (2007). nucleus. Nat. Commun. 2, 327 (2011). 30. Pittendrigh, C. S. & Daan, S. A functional analysis of circadian pacemakers in 61. Kiessling, S., Eichele, G. & Oster, H. Adrenal glucocorticoids have a key role in nocturnal rodents. III. Heavy water and constant light: homeostasis of circadian resynchronization in a mouse model of jet lag. J. Clin. Invest. 120, frequency? J. Comp. Physiol. A 106, 267–290 (1976). 2600–2609 (2010). 31. Ohta, H., Yamazaki, S. & McMahon, D. G. Constant light desynchronizes mammalian clock neurons. Nat. Neurosci. 8, 267–269 (2005). 32. Yamaguchi, Y. et al. Mice genetically deficient in vasopressin V1a and V1b Acknowledgements receptors are resistant to jet lag. Science 342,85–90 (2013). We thank A. Bradley (Wellcome Trust Sanger Institute, UK) and Y. Nakajima (National 33. Winfree, A. T. The Geometory of Biological Time, (Springer, New York, 2000). Institute of Advanced Industrial Science and Technology, Japan) for providing ROSA26-

8 NATURE COMMUNICATIONS | (2019) 10:2563 | https://doi.org/10.1038/s41467-019-10532-2 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-10532-2 ARTICLE

PBase knock-in mice and Bmal1-Eluc mice, respectively. This work was supported by the Journal peer review information: Nature Communications thanks the anonymous Core Research for Evolutional Science and Technology, Japan Science and Technology reviewer(s) for their contribution to the peer review of this work. Peer reviewer reports Agency (JPMJCR14W3), and the Ministry of Education, Culture, Sports, Science and are available. Technology of Japan (15H05933, 17H01524, and 18H04015), and AMED Project for Elucidating and Controlling Mechanisms of Ageing and Longevity (JP17gm5010002, Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in JP18gm5010002). published maps and institutional afﬁliations.

Author contributions M.D. and H.O. conceived the project and designed the research; M.D., H.S. and Y.A. Open Access This article is licensed under a Creative Commons contributed equally as ﬁrst authors who performed experiments in collaboration with Attribution 4.0 International License, which permits use, sharing, I.M., H.H., Y.T., J-M.F., Y.Y., H.K., N.K., K.Y., C.L., M.A., K.S. and H.O.; M.D. and H.O. adaptation, distribution and reproduction in any medium or format, as long as you give drafted the manuscript, supported by H.S., Y.A. and J-M.F. appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party ’ Additional information material in this article are included in the article s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the Supplementary Information accompanies this paper at https://doi.org/10.1038/s41467- article’s Creative Commons license and your intended use is not permitted by statutory 019-10532-2. regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/ Competing interests: The authors declare no competing interests. licenses/by/4.0/. Reprints and permission information is available online at http://npg.nature.com/ reprintsandpermissions/ © The Author(s) 2019

NATURE COMMUNICATIONS | (2019) 10:2563 | https://doi.org/10.1038/s41467-019-10532-2 | www.nature.com/naturecommunications 9